Studying the performance of pavement defects at different road slopes using the vibration-based method and deep machine learning

Road networks are the backbone of urban life and significantly impact the sustainability of any country's infrastructure sector. Therefore, it is necessary to maintain the condition of roads and pavements through continuous monitoring and periodic maintenance in order to achieve the highest levels of service for road users and the sustainability of their use. Pavement is the main component of road networks, providing the highest degree of comfort to drivers and roadway users when it is appropriately designed and free from defects and cracks. More clearly, defects are one of the most important factors that reduce the operational life of roads and cause economic losses to road users by causing damage to their vehicles; moreover, the damaged pavement needs frequent and long maintenance that may also drain the resources of government institutions and transport agencies. Therefore, there is a crucial need for a monitoring and follow-up system for the condition of the roads in order to identify and treat defects quickly. This study used a vibration-based system to monitor pavement conditions on several roads with different gradients. A fully electric car was used to determine the vibration values, which indicate the degree of driving comfort, to determine the spread and behaviour of defects on the pavement at multiple locations on roads with different gradients. Also, a machine learning model was applied using a "decision tree" model to identify, classify and predict defects on the pavements. The results of this study indicated that pavement defects were more prevalent in the first and last quadrants of the high-slope roads compared to the low-slope roads. The prediction model achieved accuracy in predicting the performance of defects with a rate of 94% for roads with low gradients and 90% and 86% for roads with medium and high gradients, respectively. • Using vibration-based methods to monitor pavement conditions on several roads with different gradients. • Identifying the spread and behaviour of pavement defects using a smartphone application and a fully electric passenger car. • Studying different slope gradients including low, medium, and high to understand the performance of pavement defects. • Developing a machine learning model "decision tree" to identify, classify, and predict defects on the pavements.